I was reading an article on Deep Learning and came across this term called Multi-scale Neural Network. I fully understand the concepts of convolutional neural network but it is a bit difficult to understand the multi-scale part in it. Could anyone please help me? Thanks in advance!
The term Multi-Scale used in CNNs is mainly associated with the feature extraction part. To put it simply we can say that we take an input image and resize it to various resolutions and apply a Convolution block to it to perform feature extraction.
Multi-Scale feature extraction can be done via:
- Dilated Convolutions
- Feature Pyramid Network
- Multi-Sized Filter Kernels (3x3, 5x5, 7x7, 9x9)